def _outputTablePrettyTranslate(self, page, p, table_format='github'):
tt = None
tt = TextTranslater('auto', 'en', 'us-east-1')
for table_number, table in enumerate(page.tables):
rows_list = list()
for row in table.rows:
one_row = list()
for cell in row.cells:
if cell.text != "":
one_row = one_row + [tt.getTranslation(cell.text)]
else:
one_row = one_row + [cell.text]
rows_list.append(one_row)
pretty_table = tabulate(rows_list, tablefmt=table_format)
FileHelper.writeToFile(
"{}-page-{}-table-{}-tables-pretty-translated.txt".format(
self.fileName, p, table_number), pretty_table)
def _outputFormTranslate(self, page, p):
tt = None
tt = TextTranslater('auto', 'en', 'us-east-1')
csvData = []
for field in page.form.fields:
csvItem = []
if (field.key):
csvItem.append(tt.getTranslation(field.key.text))
csvItem.append(field.key.confidence)
else:
csvItem.append("")
csvItem.append("")
if (field.value):
csvItem.append(tt.getTranslation(field.value.text))
csvItem.append(field.value.confidence)
else:
csvItem.append("")
csvItem.append("")
csvData.append(csvItem)
csvFieldNames = ['Key', 'KeyConfidence', 'Value', 'ValueConfidence']
FileHelper.writeCSV("{}-page-{}-forms-translated.csv".format(self.fileName, p),
csvFieldNames, csvData)
def generateInsights(self, insights, medicalInsights, translate, awsRegion):
print("Generating insights...")
if(not self.document.pages):
return
ta = TextAnalyzer('en', awsRegion)
tma = TextMedicalAnalyzer(awsRegion)
tt = None
if(translate):
tt = TextTranslater('auto', translate, awsRegion)
p = 1
for page in self.document.pages:
self._generateInsightsPerDocument(page, p, insights, medicalInsights, translate, ta, tma, tt)
p = p + 1